Case Study

Lenfest AI Fellow spotlight: EJ Fox

Arizona State University NEWSWELL

October 19, 2026

This profile is part of the Lenfest AI Collaborative and Fellowship Program’s 2026 Impact Report.

As an AI Fellow at Arizona State University’s NEWSWELL, EJ Fox is rethinking how small, overstretched local newsrooms can use AI to make better newspapers with fewer resources. We spoke with EJ about what he’s learned so far and what the industry gets wrong about AI.

About his role and the problem he is trying to solve

EJ’s work sits squarely at the intersection of code and journalism. He describes himself as “the computer person” in the newsroom — the one who helps reporters tell their stories with data, graphics, and now, AI.

He is the lead developer on Newskick, a tool that automates print production, built for small local newsrooms that regularly run on skeletal staff.

“Often it’s one person doing many jobs, being editor and managing the ads and doing print,” he said, referring to news operations that NEWSWELL has acquired.

“The main thing we’re trying to do is make it so that the process of creating a local newspaper is a lot easier for a smaller staff. In what ways can we make that person’s life easier so that we can have more, better local newspapers?”

What has surprised him most during the fellowship

EJ came into this work optimistic about what large language models could do for newspapers — especially because news is fundamentally about text.

“So, you’d say okay, this is a pretty good application for this technology, so let’s move around thoughts and concepts in a newspaper,” he said.

What surprised him was how limited the models still are for some of the tasks that matter most.

On the plus side, they’re reasonably good at:

Summarizing articles
Shrinking an 800-word story into a 500-word print slot

But when he tried to let the AI control the entire layout of a newspaper — deciding which article goes on the front page, where feature images should sit, or where ads should land — the results fell short both technically and editorially.

“Even with a human in the loop, we’re not seeing results that even come close to what a human page designer would do,” he said. The model makes choices that experienced editors “would never do in a million years.”

What the industry gets wrong about AI

Journalists are often too pessimistic about AI’s potential because they see the technology through the lens of automation, restructuring, and job security, he said.

EJ finds that working closely with journalists on their pain points is the solution.

“I say, ‘Hey, it looks like you’re doing a chore that you hate,’ and they’re like, ‘Yeah, I do hate this chore.’ And I’m like, ‘The robot can do that chore.’ Everyone loves having a task taken out for them, and when they see good results, it’s good.”

One lesson other newsrooms could apply tomorrow

Start extremely small. It helps both technically and with culture change.

Instead of creating a big, formal AI team or trying to overhaul workflows overnight, he advocates for:

Tiny, minimum-viable product experiments
Removing one hated chore at a time
Earning incremental buy-in from reporters and editors

What he’s paying attention to right now

EJ is closely watching the economics of AI, especially falling token prices and what they reveal about the true cost of these systems. He likens today’s AI ecosystem to the early days of Uber, when rides were heavily subsidized.

For him, the strategic question is: Do you build tools that will always require tokens, or use tokens now to build durable systems you own?

What he’s most optimistic about

EJ’s optimism comes from a sober view of the industry: In many ways, journalists have already been “pre-fired.” Newsrooms are so thinly staffed that the best use of AI isn’t in automating people out of jobs so much as filling in gaps.

He sees huge upside in freeing journalists from logistics and drudgery so they can do more of what humans excel at.

“We can just use it to make better journalism – you’ll have time to make more calls, the robot can handle the scheduling, just go talk and do the thing that only a human can do.”

The Fast Four

A book, article, or newsletter you’d recommend?
“Inventing the Future: Postcapitalism and a World Without Work,” by Nick Srnicek and Alex Williams.

What’s one thing you’re learning that has nothing to do with AI?
Becoming an amateur ham radio operator.

What local journalism challenge deserves more attention?
Existence.

One word that describes the next year of AI in news?
Reckoning. EJ anticipates a period when the industry confronts both the real capabilities and the real costs of AI — and when an increasingly skeptical public forces news organizations to be more thoughtful about how they deploy it.